A5069312410- Advanced Graph Neural Networks
- Recommender Systems and Techniques
- Nuclear physics research studies
- Topic Modeling
- Complex Network Analysis Techniques
- Opinion Dynamics and Social Influence
- AI in cancer detection
- Nuclear Physics and Applications
- Brain Tumor Detection and Classification
- Medical Image Segmentation Techniques
- Astronomical and nuclear sciences
- Multimodal Machine Learning Applications
- Nuclear reactor physics and engineering
- Head and Neck Cancer Studies
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Radiomics and Machine Learning in Medical Imaging
- Artificial Intelligence in Healthcare
- Advanced Computing and Algorithms
- Mental Health Research Topics
- Generative Adversarial Networks and Image Synthesis
- Currency Recognition and Detection
- Advanced Vision and Imaging
- Advanced Bandit Algorithms Research
- Rare-earth and actinide compounds
Wuhan Institute of Technology
2021-2024
Harbin Institute of Technology
2023-2024
Xiamen University
2023
China Institute of Atomic Energy
2006-2008
Biomedical image segmentation plays an essential role in developing computer-assisted diagnosis and treatment systems, yet it still faces numerous challenges. In the past few years, Convolutional Neural Networks (CNNs) have been successfully applied to task of biomedical segmentation. Regrettably, due locality convolution operations, these CNN-based architectures their limitations learning long-range spatial relations global context images, which might be crucial success Meanwhile, vision...
Building a universal Video-Language model for solving various video understanding tasks (e.g., text-video retrieval, question answering) is an open challenge to the machine learning field. Towards this goal, most recent works build by stacking uni-modal and cross-modal feature encoders train it with pair-wise contrastive pre-text tasks. Though offering attractive generality, resulted models have compromise between efficiency performance. They mostly adopt different architectures deal...
The continuous development of foundational models for video generation is evolving into various applications, with subject-consistent still in the exploratory stage. We refer to this as Subject-to-Video, which extracts subject elements from reference images and generates through textual instructions. believe that essence subject-to-video lies balancing dual-modal prompts text image, thereby deeply simultaneously aligning both visual content. To end, we propose Phantom, a unified framework...
Deep learning has made significant progress in computer vision, specifically image classification, object detection, and semantic segmentation. The skip connection played an essential role the architecture of deep neural networks,enabling easier optimization through residual during training stage improving accuracy testing. Many networks have inherited idea with connections for various tasks, it been standard choice designing networks. This survey provides a comprehensive summary outlook on...
The differential cross sections of quasi-elastic scattering at backward angles were measured with high precision for $^{32}\mathrm{S}+^{90,96}\mathrm{Zr}$ around the Coulomb barrier, and barrier distributions extracted from excitation functions. experimental distribution $^{32}\mathrm{S}+^{90}\mathrm{Zr}$ is well reproduced by coupled-channels calculations including low-lying quadruple octupole vibrations in $^{32}\mathrm{S}$ $^{90}\mathrm{Zr}$. However, model same coupling scheme fails to...
Predicting information cascades holds significant practical implications, including applications in public opinion analysis, rumor control, and product recommendation. Existing approaches have generally overlooked the significance of semantic topics or disregarded dissemination relations. Such models are inadequate capturing intricate diffusion process within an network inundated with diverse topics. To address such problems, we propose a neural-based model using Topic-Aware Masked Attentive...
Abstract Nasopharyngeal carcinoma (NPC) is a popular malignant tumor of the head and neck which endemic in world, more than 75% NPC patients suffer from locoregionally advanced nasopharyngeal (LA-NPC). The survival quality these depends on reliable prediction stages III IVa. In this paper, we propose two-stage framework to produce classification probabilities for predicting preprocessing MR images enhance further analysis. stage one transfer learning used improve effectiveness efficiency CNN...
An experiment of 29S bombarding on 12C target was performed to study the phenomena two‐proton halo and emission. The 1p 2p removal cross sections for are 3.15±0.32 b 1.85±0.20 b, respectively. section is abnormally large. It denotes occurs be a distribution. Among 2p‐27Si coincident events, strong correlation between two protons found. small angle indicates these might come from 2He cluster Moreover, relative momentum distribution shows locate originally at singlet 2s state emit with interaction.
Session-based Recommendation aims to reveal the item distribution patterns in anonymous session sequences. Most existing approaches model by utilizing either sequential or structural information individually absorb different pattern knowledge, which can only distinct one-sided facet of sessions, thus leading suboptimal performance. Self-supervised learning provides a natural solution as bridge fill gap between paradigms session-based recommendations, remains unexplored. In this paper, we...
Multi-modal recommendation greatly enhances the performance of recommender systems by modeling auxiliary information from multi-modality contents. Most existing multi-modal models primarily exploit multimedia propagation processes to enrich item representations and directly utilize modal-specific embedding vectors independently obtained upstream pre-trained models. However, this might be inappropriate since abundant task-specific semantics remain unexplored, cross-modality semantic gap...
Temporal Heterogeneous Networks play a crucial role in capturing the dynamics and heterogeneity inherent various real-world complex systems, rendering them noteworthy research avenue for link prediction. However, existing methods fail to capture fine-grained differential distribution patterns temporal dynamic characteristics, which we refer as spatial heterogeneity. To overcome such limitations, propose novel \textbf{C}ontrastive Learning-based \textbf{L}ink \textbf{P}rediction model,...
In order to realize solar string defect recognition, a fusion of CANNY algorithm and HOG is proposed identify defects. First, the operator used extract edge information image, then contour image after detection. By constructing feature vectors, they are input into nonlinear support vector machine (SVM) for classification. Finally, it verified in dataset, its accuracy can reach more than 90, experimental results show that method effectively improves detection efficiency
The excitation functions of elastic and quasielastic scattering at backward angles are measured for the systems 16O + 152Sm , 6,7Li 208Pb. barrier distributions extracted from these compared with corresponding fusion distributions. Except some details, derived data elastic/quasielastic almost same tightly bound reaction systems. For weakly projectile, obviously different However, plus breakup as one complete data. This result means that distribution not only bears information nuclear...